Uridylate-specific 3′–5′-Exoribonucleases Involved in Uridylate-deletion RNA Editing in Trypanosomatid Mitochondria*

In kinetoplastid protists, maturation of mitochondrial pre-mRNAs involves the insertion and deletion of uridylates (Us) within coding regions, as specified by mitochondrial DNA-encoded guide RNAs. U-deletion editing involves endonucleolytic cleavage of the pre-mRNA at the editing site followed by U-specific 3′–5′-exonucleolytic removal of nonbase-paired Us prior to ligation of the two mRNA cleavage fragments. We showed previously that an exonuclease/endonuclease/phosphatase (EEP) motif protein from Leishmania major, designated RNA editing exonuclease 1 (REX1) (Kang, X., Rogers, K., Gao, G., Falick, A. M., Zhou, S.-L., and Simpson, L. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 1017–1022), exhibits 3′–5′-exonuclease activity. Two EEP motif proteins have also been identified in the Trypanosoma brucei editing complex. TbREX1 is a homologue of LmREX1, and TbREX2 shows homology to another editing protein in L. major, which lacks the EEP motif (LmREX2*). Here we have expressed the T. brucei EEP motif proteins in insect cells and purified them to homogeneity. We showed that these are U-specific 3′–5′-exonucleases that are inhibited by base pairing of 3′ Us. The recombinant EEP motif alone also showed 3′–5′ U-specific exonuclease activity, and mutations of the REX EEP motifs greatly reduced exonuclease activity. The absence of enzymatic activity in LmREX2* was confirmed with a purified recombinant protein. We showed that pre-cleaved U-deletion editing could be reconstituted with either TbREX1 or TbREX2 in combination with either RNA ligase, LmREL1, or LmREL2. Down-regulation of TbREX2 expression by conditional RNA interference had little effect on parasite viability or sedimentation of the L-complex, suggesting either that TbREX2 is inactive in vivo or that TbREX1 can compensate for the loss of TbREX2 function in down-regulated cells.

In kinetoplastid protists, maturation of mitochondrial pre-mRNAs involves the insertion and deletion of uridylates (Us) within coding regions, as specified by mitochondrial DNA-encoded guide RNAs. U-deletion editing involves endonucleolytic cleavage of the pre-mRNA at the editing site followed by U-specific 3-5-exonucleolytic removal of nonbase-paired Us prior to ligation of the two mRNA cleavage fragments. We showed previously that an exonuclease/endonuclease/phosphatase (EEP) motif protein from Leishmania major, designated RNA editing exonuclease 1 (REX1) (Kang, X., Rogers, K., Gao A. 102, 1017-1022), exhibits 3-5-exonuclease activity. Two EEP motif proteins have also been identified in the Trypanosoma brucei editing complex. TbREX1 is a homologue of LmREX1, and TbREX2 shows homology to another editing protein in L. major, which lacks the EEP motif (LmREX2*). Here we have expressed the T. brucei EEP motif proteins in insect cells and purified them to homogeneity. We showed that these are U-specific 3-5-exonucleases that are inhibited by base pairing of 3 Us. The recombinant EEP motif alone also showed 3-5 U-specific exonuclease activity, and mutations of the REX EEP motifs greatly reduced exonuclease activity. The absence of enzymatic activity in LmREX2* was confirmed with a purified recombinant protein. We showed that pre-cleaved U-deletion editing could be reconstituted with either TbREX1 or TbREX2 in combination with either RNA ligase, LmREL1, or LmREL2. Down-regulation of TbREX2 expression by conditional RNA interference had little effect on parasite viability or sedimentation of the L-complex, suggesting either that TbREX2 is inactive in vivo or that TbREX1 can compensate for the loss of TbREX2 function in down-regulated cells.
The majority of mitochondrial pre-mRNAs in kinetoplastid protists are edited through the insertion and deletion of uridy-late residues (Us). 3 The editing process is directed by guide RNAs (gRNAs) that initially anneal to the pre-edited mRNA at the first downstream editing site via a region of complementarity termed the "anchor sequence." An editing cycle involves an enzyme cascade (2) initiated by specific RNA editing endonucleases (REN), one of which (REN1) cleaves the mRNA immediately 5Ј to the mRNA/gRNA duplex at U-deletion sites and the other (REN2) cleaves at U-insertion sites (3)(4)(5). The upstream gRNA sequence directs the modification of the 3Ј end of the 5Ј mRNA fragment by either the insertion of complementary Us by an RNA editing terminal uridylyltransferase (RET2) or by the removal of nonbase-paired Us by an RNA editing exonuclease (REX). Ligation of the mRNA fragments by an RNA editing ligase (REL) terminates the single site editing cycle. Editing then proceeds to the adjacent upstream site until all sites mediated by the single gRNA are edited resulting in an edited mRNA/gRNA duplex. Multiple overlapping gRNAs are required for extensive editing (pan-editing). Editing is catalyzed by a ribonucleoprotein complex (editosome) consisting of the RNA ligase-containing complex (L-complex) together with several additional complexes involved with RNA binding and 3Ј U addition to the gRNAs linked by bound RNA (6).
The current understanding of U-insertion/deletion RNA editing encompasses data derived from both Trypanosoma brucei and Leishmania tarentolae model systems. Up to 20 L-complex components have been identified experimentally in T. brucei (7) and L. tarentolae (8), and orthologues of L-complex components have been identified in the Leishmania major genome (7,8). Enzymatic activities consistent with the enzyme cascade model have been confirmed in vitro for several components of the L-complex, including REL1, REL2 (9 -11), RET2 (12,13), and REN1 (5). Identified activities were similar in orthologues from both Leishmania and Trypanosoma. REX1 activity has also been confirmed in vitro but only with the L. major orthologue (1).
LmREX1 was first identified as a candidate exonuclease by the presence of a conserved exonuclease/endonuclease/ phosphatase (EEP, Pfam accession number PF03372) domain (6). We subsequently found recombinant LmREX1 to be a 3Ј-5Ј U-specific exonuclease (1). However, the T. brucei L-complex contains a second EEP domain protein (14 -17), which we designated REX2 for its putative function (1). TbREX2 is a component of the REL1 subcomplex of the L-complex that is competent for "pre-cleaved" deletion editing in vitro (18). Surprisingly, the Leishmania orthologue of TbREX2 lacks an EEP domain, suggesting an absence of exonuclease activity, and we have named the Leishmania orthologue REX2*. Here we further investigate the kinetoplastid REX proteins by purification and enzymatic characterization of recombinant REX proteins from T. brucei, and we demonstrate for the first time a difference in enzymatic content of the Leishmania and Trypanosoma L-complexes.
Cytosolic Expression of the TbREX1 EEP Domain-A TbREX1 gene fragment coding for amino acids 606 -902 was PCR-amplified using the primer pair 5Ј-GGATCCATGGC-AAAAGAGGTT-3Ј and 5Ј-AAGCTTCTATCTAGAAAG-GGCACCAATAATCTGAACTTCCG-3Ј and cloned into pCR2.1-TOPO. A modified TAP tag containing a protein C epitope instead of cbp was PCR-amplified from the pC-PTP-NEO vector (19) using the primers 5Ј-TCTAGAGGCTCCGG-CTCCATGGGAAGATCAGGTGGAT-3Ј and 5Ј-GCGGCC-GCTCAGGTTGACTTCCC-3Ј and after digestion with XbaI and NotI was inserted into the corresponding sites of the LtREL1-TAP (6) episomal expression vector, replacing the cbp tag. An internal BamHI site was removed from the protein C epitope using the QuickChange site-directed mutagenesis kit and the primer 5Ј-CCGGCTCCATGGAAGAAGATCAGGT-GGACCCTCGTCTTATTGATG-3Ј. The TbREX1-(606 -902) sequence was released from pCR2.1-TOPO with BamHI and XbaI and inserted into the corresponding sites of the modified TAP vector. L. tarentolae UC strain cells were maintained in BHI media (Difco) supplemented with 10 mg/ml hemin-HCl. Cells were electroporated with the TAP vector and selected for retention of plasmid by plating on agarose in the presence of 200 g/ml G418. Transformed cells were maintained in 100 g/ml G418 (20). Affinity purification of TbREX1-(606 -902)protein C was performed as described previously (19).
Tandem Affinity Purification of LmREX2*-L-complex containing TAP-tagged LmREX2* was purified from L. tarentolae. LmREX2* released from pCR2.1-TOPO-Lm-REX2* with BamHI and XbaI and inserted into the corresponding sites of LtREL1-TAP (6) to generate LmREX2*-TAP vector for transformation of cells was conducted as described above. Tandem affinity purification of L-complex was performed as described previously (6).
Cloning of RNAi Vector and Down-regulation of TbREX2-A 500-bp fragment of the TbREX2 gene was PCR-amplified using the primers 5Ј-GGATCCAAGCTTATGTTGCGCCGCAGT-CGCTT-3Ј and 5Ј-GGATCCTCTAGACGAGGTCAAGCA-TCCACGCA-3Ј and cloned into pCR2.1-TOPO. The gene fragment was released first with HindIII and XbaI and subsequently with BamHI for insertion into the corresponding sites of the pLew100-HX-GFP vector as inverted copies separated by a spacer element (21). The fragment was also released with BamHI and HindIII and inserted into the corresponding sites of the p2T7 Ti -177 vector (22). RNAi vectors were linearized with NotI and electroporated into strain 29-13 T. brucei procyclic form cells that express T7 RNA polymerase and tetracycline repressor. Procyclic cells were maintained in SDM79 media supplemented with 15% fetal bovine serum, 15 g/ml G418, and 50 g/ml hygromycin. After electroporation, cultures were selected for genomic integration of the construct at the rDNA spacer (pLew100-HX-GFP) or the 177-bp satellite (p2T7 Ti -177) by plating on agarose in the presence of 2.5 g/ml phleomycin (23). Transcription of TbREX2 dsRNA was induced by addition of 1 g/ml tetracycline.
For growth curve analysis, cells were maintained in log phase by dilution to 10 6 cells/ml every 48 h. Mitochondria were isolated from log phase cultures as described (24). Extracts from 100 mg of purified mitochondria were prepared by sonication in 50 mM HEPES, pH 7.9, 60 mM KCl, 10 mM MgCl 2 , and 0.5% Nonidet P-40. Extracts were clarified by centrifugation for 30 min at 70,000 rpm in a Beckman TLA-100.4 rotor and sedimented in a 10 -30% glycerol gradient for 20 h at 30,000 rpm in a Beckman SW41Ti rotor. Gradient fractions were auto-adenylated for 20 min at 27°C with 5 Ci of [␣-32 P]ATP per 10 l and resolved by SDS-PAGE on 8 -16% acrylamide NOVEX gels (Invitrogen) prior to transfer to Protran BA83 nitrocellulose (Schleicher & Schuell) for immunodetection or autoradiography. TbREX2 mRNA levels were determined by reverse transcription-PCR analysis. Total RNA was purified and DNase treated using the RNeasy mini kit (Qiagen). SuperScript II reverse transcriptase (Invitrogen) was used for cDNA synthesis. The primer pairs 5Ј-TGATCATGTCGCCGTCGACG-3Ј, 5Ј-TTGAAGTTCACGCCCAGGCC-3Ј; and 5Ј-TGACTACG-TTGCGGTGGACGCC-3Ј, 5Ј-CTCGCGGTCGCGCTTTG-AGA-3Ј were used to PCR-amplify regions of TbREX2 and TbREX1 respectively.
Kinetic Analysis of TbREX1 and TbREX2 Activities-rTbREX1 and rTbREX2 were incubated with 5Ј end-labeled 1U substrate RNA in standard reaction buffer. To determine the linear range of product accumulation, time course reactions at various REX protein concentrations were monitored by denaturing PAGE. Kinetic parameters were determined by incubating 22 fmol of rTbREX1 or 54 fmol of rTbREX2 in 10-l reactions with varying concentrations of 1U substrate for 10 and 15 min, respectively. Substrate concentration ranged from 50 nM to 1.5 M. The percent of substrate cleavage visualized by PhosphorImager was quantified using ImageQuant (version 5.2, GE Healthcare) and used to calculate enzyme velocity (fmol/min). Kinetic parameters were calculated from nonlinear regression of velocity versus substrate concentration plots using GraphPad Prism (version 4.01, GraphPad Software, Inc.).
Reactions contained 1 pmol of kinase-labeled 5Ј UU fragment, 2 pmol of 3Ј fragment, and 4 pmol of a single gRNA. RNAs were annealed by heating to 70°C and slowly cooling to 4°C prior to addition of reaction buffer. Recombinant TbREL1 and TbREL2 purified from Sf9 cells were added to some reactions (9).

Expression and Purification of the T. brucei and L. major REX
Proteins-PFAM analysis of the TbREX1, TbREX2, and LmREX1 amino acid sequences indicated the presence of the EEP motif in all three sequences (Fig. 1A), as demonstrated previously. Also as noted previously, the LmREX2 homologue sequence lacks any vestige of this motif, leading us to use the LmREX2* nomenclature for this protein. As discussed in the Introduction, this finding raises a paradox because the REL1 subcomplex of the L-complex in T. brucei and L. tarentolae, which has been proposed to be mainly involved in U-deletion editing, appears to contain only REX2 (6,18). To address the possible functional role of LmREX2*, we decided to analyze a recombinant LmREX2* protein in addition to recombinant TbREX1 and TbREX2 proteins. The purification to homogeneity of insect cell-expressed TAP-tagged recombinant TbREX1, TbREX2, and LmREX2* proteins by successive affinity and chromatographic steps is shown in Fig. 1B.
Exonuclease Activity of Recombinant REX Proteins-The recombinant TbREX1 and TbREX2 proteins were found to exhibit a robust 3Ј-5Ј-exonuclease activity (Fig. 2, A and B). This activity was specific for 3Ј uridylates, and no degradation was observed with RNAs terminating in 3Ј A, G, or C residues. This exouridylase activity appeared distributive in nature. A processive type activity was observed using only affinity-purified proteins (data not shown). The rTbREX activity was dependent on magnesium cation and was sensitive to chelating agents (Fig. 2C).

JOURNAL OF BIOLOGICAL CHEMISTRY 29075
observed even when the rLmREX2* concentration was increased by 2 orders of magnitude versus the concentrations of rTbREX1 and rTbREX2 used previously. The presence of LtREX2* in the L-complex was demonstrated by transfecting L. tarentolae cells with TAP-tagged LmREX2* and showing that the TAP pulldown had an identical polypeptide profile as L-complex isolated by use of known L-complex components (data not shown).
Down-regulation of TbREX2 Expression Has a Minor Effect on Cell Growth-TbREX2 gene expression was down-regulated by tetracycline-inducible RNA interference. A procyclic stage T. brucei cell line was generated in which transcription of stemloop RNA from inverted copies of a 500-bp TbREX2 gene fragment was driven by a tetracycline-inducible PARP promoter. Induction of RNAi for 3 days resulted in reduction of TbREX2 mRNA levels as detected by reverse transcription-PCR from total RNA (Fig. 3A, right panel). Loss of TbREX2 expression had little effect on parasite viability and growth (Fig. 3A, left panel).
However, after 5 days of induction a slight but consistent reduction in growth rate was observed. An alternate RNAi cell line expressing TbREX2 dsRNA from opposing T7 RNA polymerase promoters exhibited no growth phenotype after RNAi induction (data not shown). This behavior contrasts to the previously reported strong effect of down-regulation of TbREX1 expression on cell growth (1). A majority of proteins were expressed in insect cells as TAP tag fusions. Proteins were purified by sequential affinity, cation exchange, and gel filtration steps. rTbREX1-(606 -902) was expressed in the cytosol of L. tarentolae as a modified TAP tag fusion containing protein C instead of cbp. rTbREX1-(606 -902) was purified by sequential affinity chromatography. B, purity was assessed by SYPRO Ruby staining after SDS-PAGE in 8 -16% acrylamide gels. Molecular weight markers are indicated for each gel. FIGURE 2. rTbREX1 and rTbREX2 are 3-5 U-specific exonucleases, whereas rLmREX2* appears inactive. rTbREX1 (2.8 nM) (A) or rTbREX2 (5.4 nM) (B) were incubated with 5Ј 32 P-labeled single-stranded RNA substrates 3Ј terminating in 6 Us, As, Gs, or Cs. At the indicated time points, reaction products were separated by 8 M urea, 15% acrylamide denaturing PAGE, and dried gels were visualized with PhosphorImager screens. Mock reactions (M) did not include enzyme. C, rTbREX1 (left panel) and rTbREX2 (right panel) require magnesium for activity against 1U RNA with a single 3Ј uridylate. D, rLmREX2* was assayed after 32 min of incubation for nuclease activity at increasing protein concentrations.
The effect of TbREX2 down-regulation on L-complex stability was assayed by sedimentation of mitochondrial lysate before and after 9 days of RNAi induction. Only minor changes in L-complex sedimentation were apparent (Fig. 3B), suggesting that TbREX2 is not required to maintain the stability or the structure of the L-complex. In addition, no major differences were detected among the distribution of the L-complex components LC1 (MP81), LC4 (MP63), LC7b (MP42), and REL2. The increased level of REL1 detected in fraction three after RNAi is consistent with previous findings that suggest the presence of a REL1 subcomplex consisting of REX2, REL1, and LC4 (18). REX2 may help stabilize the interaction of REL1 with the L-complex.
REX EEP Domains Are Required for Exonuclease Activity-The EEP domains from diverse proteins share several highly conserved amino acids (25), which are also conserved in the TbREX1 and TbREX2 EEP domains (Fig. 4A) (26,27). To confirm that the ribonuclease activity of both TbREX1 and TbREX2 is because of the EEP domains, TbREX proteins mutated in the conserved EEP motif amino acids were tested for exonuclease activity in vitro. Two amino acids were mutated in both TbREX1 (D881A,H882N), and TbREX2 (D896A,H897N), and the mutant proteins were purified after overexpression in insect cells (Fig. 1B). The exonuclease activity of the mutant proteins was evaluated with a single-stranded RNA substrate terminating in a single 3Ј uridylate. Time course analysis of both wild type and mutant protein reactions showed decreased activity of mutant versus wild type REX proteins (Fig. 4B).
The role of the EEP motif in catalysis was further examined through expression of the TbREX1 C-terminal EEP domain. A truncated TbREX1 peptide containing amino acids 606 -902 was purified after overexpression in the cytosol of L. tarentolae. rTbREX1-(606 -902) was sufficient for 3Ј-5Ј-exonuclease activity, and the exonuclease activity was U-specific (Fig. 4C). Our previous observation that, during the purification of LmREX1, C-terminal proteolytic fragments co-fractionated with a nonspecific exonuclease activity (1) could represent contamination with nonspecific nucleases, conformational changes induced by proteolysis, or authentic differences in the structural stability and determinants of specificity for EEP domains from L. major and T. brucei.
Kinetic Analysis of REX 3Ј-5Ј-Exonuclease Activity-Steady state kinetics of the TbREX exonuclease reactions were determined using a single-stranded RNA substrate terminating in a single 3Ј U residue. The rTbREX1 and rTbREX2 reactions yielded similar apparent K m values of 536.1 Ϯ 84.99 and 224.5 Ϯ 25.83 nM, indicating a high affinity of these enzymes for the RNA substrate (Fig. 5, A and B). The catalytic efficiencies (k cat /K m ) were 1.7 ϫ 10 4 and 6.2 ϫ 10 3 s Ϫ1 M Ϫ1 , respectively. The exonuclease efficiency of TbREX1 was 2.7-fold higher than that of TbREX2; however, the similarity of the two values suggests that both enzymes may exhibit functionally relevant catalytic activity in vivo.
REX1 and REX2 Exouridylase Activity Is Affected by gRNA Sequence-Although the TbREX enzymes exhibit similar activities against simple substrates in vitro, the REX proteins may exhibit alternate specificities on complex substrates such as editing sites in vivo. To simulate substrate conditions at U-deletion editing sites, rTbREX enzymes were tested for activity against pre-cleaved RNA substrates that resemble deletion editing sites after an initial endonucleolytic cleavage step (28). Both enzymes were capable of 3Ј-5Ј-exonucleolytic removal of unpaired Us from the 3Ј end of the 5Ј mRNA fragment (Fig. 6A). Both enzymes also exhibited a preference for single-stranded 3Ј Us; trimming of the 3Ј-terminal Us stopped to a large extent at duplex regions generated by complementary gRNAs (Fig. 6, A  and B). The partial degradation of base-paired 3Ј Us may be due to limited breathing of the annealed substrate RNAs.
Both REX1 and REX2 Remove Us from Pre-cleaved Deletion Substrates and Can Be Used to Reconstitute Pre-cleaved U-deletion RNA Editing in Vitro-We have shown previously that pre-cleaved deletion editing can be reconstituted using the Leishmania proteins, rLmREX1 and rLmREL1 (1). Full round single site deletion editing can be reconstituted by the further addition of the U-deletion site-specific endonuclease, rLmMP90 (rLmREN2) (5). Recombinant TbREX1 and TbREX2 were combined with recombinant TbREL1 and TbREL2 RNA ligases (9) that were purified from overexpression in insect cells and incubated with the Ϫ2U pre-cleaved substrate. Any combination of REX and REL proteins was sufficient to generate fully edited products (Fig. 6B). This is in marked contrast to the Leishmania system in which the rLmREX1 and rLmREL1 com- bination was competent for editing but the rLmREX1 and rLm-REL2 combination was not (1).

DISCUSSION
We have demonstrated that TbREX1 and TbREX2 are singlestranded U-specific 3Ј-5Ј-exoribonucleases. This is consistent with our previous characterization of the Leishmania EEP motif protein, LmREX1 (1), and with the known specificity of U-deletion exouridylase activity in mitochondrial lysates. Mutations of conserved residues in the TbREX1 and TbREX2 EEP domains greatly reduced exonuclease activity, indicating that these domains are required for exonuclease catalysis. This was confirmed by showing that a truncated TbREX1 peptide containing only the EEP domain is sufficient for U-specific exonuclease activity. This is consistent with our finding that recombinant LmREX2*, which lacks this motif, has no detectable exonuclease activity. Together, these results represent the first demonstrated difference in enzymatic content of the Leishmania and Trypanosoma L-complexes, raising the possibility that other differences may exist.
The kinetoplastid REX1 and REX2 proteins are the only known 3Ј-5Ј U-specific exonucleases. No orthologues of these enzymes have been identified outside the order Kinetoplastida; however, other EEP motif proteins may provide some insight into REX function. Although nucleotide-specific exonuclease activities are rare, the EEP protein family members CCR4 and Nocturnin, two  3Ј-5Ј RNA deadenylases, may provide the closest functional models for the REX proteins (29 -31). CCR4 exhibits both distributive and processive activity in a substrate-dependent fashion, possibly as a function of mRNA length (32). We showed that the highly purified T. brucei REX proteins exhibit a distributive activity, but proteins purified only by affinity chromatography show a processive activity (data not shown).
Homology modeling has suggested that REX EEP domains may adopt a four-layer ␣/␤ fold similar to that of EEP proteins with known structures (27), and consequently, TbREX enzymes may utilize a catalytic mechanism similar to those described for other EEP proteins. However, although conserved residues that characterize the EEP motif are located in a single active site, distinct catalytic mechanisms have been proposed for various EEP model proteins (33,34). Our double amino acid mutations of TbREX1 (D881A,H882N), and TbREX2 (D896A,H897N) were based on functional analyses of the EEP motif protein, APE1, an apurinic endonuclease involved in base excision repair. Single amino acid mutations of equivalent residues in APE1 have demonstrated that these residues are important for catalytic activity (35,36), and crystal structures of APE1 have confirmed the localization of these residues at the active site (33,37). Further mutational analysis may provide some insight into the exact catalytic mechanism of REX proteins.
A 5-10 S subcomplex containing TbREL1, TbMP63, and TbREX2 has been identified by gradient sedimentation of TbREL1 TAP-purified L-complex and found to be capable of pre-cleaved deletion editing (18). A similar subcomplex containing LtREX2*, LtREL1, and LtLC4 has been identified in L. tarentolae (6). However, the lack of enzymatic activity of LmREX2* suggests that that in Leishmania this subcomplex may not be involved in U-deletion editing. The role of Leishmania REX2* remains unclear. One possibility is that LmREX2* may simply play a structural role, perhaps stabilizing interactions within the REL1 subcomplex. Presumably, REX1 acts as the primary U-deletion editing exonuclease in the Leishmania system. Interactions between REX1 and other L-complex components have yet to be identified. It remains possible that the Leishmania REX1 protein may functionally interact with the REL1 sub-complex in vivo or substitute for LtREX2* in a portion of the total subcomplexes.
Analysis of the mRNA cleavage step of editing in gradientpurified 20 S fractions first indicated that separate endonucleases act at insertion and deletion editing sites (38). Endonuclease activity preferentially at U-deletion editing sites of the L-complex protein, TbREN1, have been confirmed in vitro with recombinant REN1 (5) and in vivo through mutation of the TbREN1 RNase III motif (3). A U-insertion site preference of another RNase III endonuclease, TbREN2, has been identified in vivo (4). TbREN1 and TbREN2 appear to be mutually exclusive in L-complexes isolated by either TbREN1 TAP or TbREN2 TAP (39). However, although TbREX2 was present in both L-complexes, TbREX1 was only identified in the TbREN1 complex (39).
The similarity of rTbREX1 and rTbREX2 activities in vitro suggests that these enzymes may act redundantly in vivo and may both play catalytic roles in U-deletion editing. This is supported by the successful reconstitution of pre-cleaved deletion editing with either TbREX enzyme. The minimal phenotype of down-regulated TbREX2 cells and the slow growth phenotype of down-regulated TbREX1 cells (1) are consistent with a limited ability of REX enzymes to act redundantly in vivo. The stronger growth phenotype resulting from down-regulation of TbREX1 (1) suggests that TbREX2 may not fully substitute for TbREX1 function. However, RNAi down-regulation of TbREX1 expression produces a shift in sedimentation value of the L-complex greater than that observed upon TbREX2 downregulation. We cannot discount the possibility that the slow growth phenotype of down-regulated TbREX1 cells is because of loss of other L-complex components rather than an inability of TbREX2 to compensate for loss of TbREX1. The REX pro-FIGURE 6. rTbREX1 and rTbREX2 exonuclease activities are gRNA-directed. A, REX proteins were incubated with pre-cleaved deletion editing substrates with different gRNA sequences as shown in the diagram. Basepairing between guiding nucleotides of the gRNA and 3Ј Us reduces TbREX activity. B, reconstitution of pre-cleaved U deletion editing with rTbREX and rTbREL proteins. Ϫ1 U and Ϫ2 U fully edited products were generated from the Ϫ2U-deletion substrate with any combination of rTbREX and rTbREL proteins. OCTOBER 5, 2007 • VOLUME 282 • NUMBER 40 teins might also have distinct in vivo functions in T. brucei, including specificity for specific deletion substrates or trimming of excess 3Ј Us from the 5Ј cleavage fragment possibly added by RET1 or RET2 terminal uridylyltransferase activity at U-deletion or U-insertion sites (40 -42).

Editing Exonucleases
The presence of several homologues of editing proteins, one of which is required and the other nonessential, has been also demonstrated for the RNA ligases REL1 and REL2. It is not clear if this organizational trend in editing proteins represents a functional organization or is an evolutionary artifact.